from typing import List, Optional, Tuple, Union import torch import torch.nn as nn from PIL import Image from torch.nn import CrossEntropyLoss from transformers import ( AutoConfig, AutoModelForCausalLM, MixtralConfig, MixtralForCausalLM, MixtralModel, ) from transformers.cache_utils import Cache, DynamicCache from transformers.modeling_outputs import CausalLMOutputWithPast, MoeCausalLMOutputWithPast from ..vita_arch import VITAMetaForCausalLM, VITAMetaModel def load_balancing_loss_func( gate_logits: torch.Tensor, num_experts: torch.Tensor = None, top_k=2, attention_mask: Optional[torch.Tensor] = None, ) -> float: r""" Computes auxiliary load balancing loss as in Switch Transformer - implemented in Pytorch. See Switch Transformer (https://arxiv.org/abs/2101.03961) for more details. This function implements the loss function presented in equations (4) - (6) of the paper. It aims at penalizing cases where the routing between experts is too unbalanced. Args: gate_logits (Union[`torch.Tensor`, Tuple[torch.Tensor]): Logits from the `gate`, should be a tuple of model.config.num_hidden_layers tensors of shape [batch_size X sequence_length, num_experts]. attention_mask (`torch.Tensor`, None): The attention_mask used in forward function shape [batch_size X sequence_length] if not None. num_experts (`int`, *optional*): Number of experts Returns: The auxiliary loss. """ if gate_logits is None or not isinstance(gate_logits, tuple): return 0 if isinstance(gate_logits, tuple): compute_device = gate_logits[0].device concatenated_gate_logits = torch.cat( [layer_gate.to(compute_device) for layer_gate in gate_logits], dim=0 ) routing_weights = torch.nn.functional.softmax(concatenated_gate_logits, dim=-1) _, selected_experts = torch.topk(routing_weights, top_k, dim=-1) expert_mask = torch.nn.functional.one_hot(selected_experts, num_experts) if attention_mask is None: # Compute the percentage of tokens routed to each experts tokens_per_expert = torch.mean(expert_mask.float(), dim=0) # Compute the average probability of routing to these experts router_prob_per_expert = torch.mean(routing_weights, dim=0) else: batch_size, sequence_length = attention_mask.shape num_hidden_layers = concatenated_gate_logits.shape[0] // (batch_size * sequence_length) # Compute the mask that masks all padding tokens as 0 with the same shape of expert_mask expert_attention_mask = ( attention_mask[None, :, :, None, None] .expand((num_hidden_layers, batch_size, sequence_length, top_k, num_experts)) .reshape(-1, top_k, num_experts) .to(compute_device) ) # Compute the percentage of tokens routed to each experts tokens_per_expert = torch.sum( expert_mask.float() * expert_attention_mask, dim=0 ) / torch.sum(expert_attention_mask, dim=0) # Compute the mask that masks all padding tokens as 0 with the same shape of tokens_per_expert router_per_expert_attention_mask = ( attention_mask[None, :, :, None] .expand((num_hidden_layers, batch_size, sequence_length, num_experts)) .reshape(-1, num_experts) .to(compute_device) ) # Compute the average probability of routing to these experts router_prob_per_expert = torch.sum( routing_weights * router_per_expert_attention_mask, dim=0 ) / torch.sum(router_per_expert_attention_mask, dim=0) overall_loss = torch.sum(tokens_per_expert * router_prob_per_expert.unsqueeze(0)) return overall_loss * num_experts def custom_forward( self, input_ids: torch.LongTensor = None, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.LongTensor] = None, past_key_values: Optional[List[torch.FloatTensor]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, output_router_logits: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[Tuple, MoeCausalLMOutputWithPast]: r""" Args: labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*): Labels for computing the masked language modeling loss. Indices should either be in `[0, ..., config.vocab_size]` or -100 (see `input_ids` docstring). Tokens with indices set to `-100` are ignored (masked), the loss is only computed for the tokens with labels in `[0, ..., config.vocab_size]`. Returns: Example: ```python >>> from transformers import AutoTokenizer, MixtralForCausalLM >>> model = MixtralForCausalLM.from_pretrained("mistralai/Mixtral-8x7B-v0.1") >>> tokenizer = AutoTokenizer.from_pretrained("mistralai/Mixtral-8x7B-v0.1") >>> prompt = "Hey, are you conscious? Can you talk to me?" >>> inputs = tokenizer(prompt, return_tensors="pt") >>> # Generate >>> generate_ids = model.generate(inputs.input_ids, max_length=30) >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you." ```""" output_attentions = ( output_attentions if output_attentions is not None else self.config.output_attentions ) output_router_logits = ( output_router_logits if output_router_logits is not None else self.config.output_router_logits ) output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) outputs = self.model( input_ids=input_ids, attention_mask=attention_mask, position_ids=position_ids, past_key_values=past_key_values, inputs_embeds=inputs_embeds, use_cache=use_cache, output_attentions=output_attentions, output_hidden_states=output_hidden_states, output_router_logits=output_router_logits, return_dict=return_dict, ) hidden_states = outputs[0] logits = self.lm_head(hidden_states) # logits = logits.float() loss = None if labels is not None: # Shift so that tokens < n predict n shift_logits = logits[..., :-1, :].contiguous() shift_labels = labels[..., 1:].contiguous() # Flatten the tokens loss_fct = CrossEntropyLoss() shift_logits = shift_logits.view(-1, self.config.vocab_size) shift_labels = shift_labels.view(-1) # Enable model parallelism shift_labels = shift_labels.to(shift_logits.device) loss = loss_fct(shift_logits, shift_labels) aux_loss = None if output_router_logits: aux_loss = load_balancing_loss_func( outputs.router_logits if return_dict else outputs[-1], self.num_experts, self.num_experts_per_tok, attention_mask, ) if labels is not None: loss += self.router_aux_loss_coef * aux_loss.to( loss.device ) # make sure to reside in the same device if not return_dict: output = (logits,) + outputs[1:] if output_router_logits: output = (aux_loss,) + output return (loss,) + output if loss is not None else output return MoeCausalLMOutputWithPast( loss=loss, aux_loss=aux_loss, logits=logits, past_key_values=outputs.past_key_values, hidden_states=outputs.hidden_states, attentions=outputs.attentions, router_logits=outputs.router_logits, ) MixtralForCausalLM.forward = custom_forward class VITAMixtralConfig(MixtralConfig): model_type = "vita-mixtral" class VITAMixtralModel(VITAMetaModel, MixtralModel): config_class = VITAMixtralConfig def __init__(self, config: MixtralConfig): super(VITAMixtralModel, self).__init__(config) class VITAMixtralForCausalLM(MixtralForCausalLM, VITAMetaForCausalLM): config_class = VITAMixtralConfig def __init__(self, config): super(MixtralForCausalLM, self).__init__(config) self.model = VITAMixtralModel(config) self.vocab_size = config.vocab_size self.lm_head = nn.Linear(config.hidden_size, config.vocab_size, bias=False) self.router_aux_loss_coef = config.router_aux_loss_coef self.num_experts = config.num_local_experts self.num_experts_per_tok = config.num_experts_per_tok # Initialize weights and apply final processing self.post_init() def get_model(self): return self.model def forward( self, input_ids: torch.LongTensor = None, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.LongTensor] = None, past_key_values: Optional[List[torch.FloatTensor]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, images: Optional[torch.FloatTensor] = None, audios: Optional[dict] = None, sf_masks: Optional[torch.Tensor] = None, output_router_logits: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[Tuple, CausalLMOutputWithPast]: if inputs_embeds is None: ( input_ids, position_ids, attention_mask, past_key_values, inputs_embeds, labels, ) = self.prepare_inputs_labels_for_multimodal( input_ids, position_ids, attention_mask, past_key_values, labels, images, audios, sf_masks ) return super().forward( input_ids=input_ids, attention_mask=attention_mask, position_ids=position_ids, past_key_values=past_key_values, inputs_embeds=inputs_embeds, labels=labels, use_cache=use_cache, output_attentions=output_attentions, output_hidden_states=output_hidden_states, output_router_logits=output_router_logits, return_dict=return_dict, ) def prepare_inputs_for_generation_original( self, input_ids, past_key_values=None, attention_mask=None, inputs_embeds=None, output_router_logits=False, **kwargs, ): # Omit tokens covered by past_key_values if past_key_values is not None: if isinstance(past_key_values, Cache): cache_length = past_key_values.get_seq_length() past_length = past_key_values.seen_tokens max_cache_length = past_key_values.get_max_length() else: cache_length = past_length = past_key_values[0][0].shape[2] max_cache_length = None # Keep only the unprocessed tokens: # 1 - If the length of the attention_mask exceeds the length of input_ids, then we are in a setting where # some of the inputs are exclusively passed as part of the cache (e.g. when passing input_embeds as # input) if attention_mask is not None and attention_mask.shape[1] > input_ids.shape[1]: input_ids = input_ids[:, -(attention_mask.shape[1] - past_length) :] # 2 - If the past_length is smaller than input_ids', then input_ids holds all input tokens. We can discard # input_ids based on the past_length. elif past_length < input_ids.shape[1]: input_ids = input_ids[:, past_length:] # 3 - Otherwise (past_length >= input_ids.shape[1]), let's assume input_ids only has unprocessed tokens. else: remove_prefix_length = input_ids.shape[1] - 1 input_ids = input_ids[:, remove_prefix_length:] # If we are about to go beyond the maximum cache length, we need to crop the input attention mask. if ( max_cache_length is not None and attention_mask is not None and cache_length + input_ids.shape[1] > max_cache_length ): attention_mask = attention_mask[:, -max_cache_length:] position_ids = kwargs.get("position_ids", None) if attention_mask is not None and position_ids is None: # create position_ids on the fly for batch generation position_ids = attention_mask.long().cumsum(-1) - 1 position_ids.masked_fill_(attention_mask == 0, 1) if past_key_values: position_ids = position_ids[:, -input_ids.shape[1] :] # if `inputs_embeds` are passed, we only want to use them in the 1st generation step if inputs_embeds is not None and past_key_values is None: model_inputs = {"inputs_embeds": inputs_embeds} else: model_inputs = {"input_ids": input_ids} model_inputs.update( { "position_ids": position_ids, "past_key_values": past_key_values, "use_cache": kwargs.get("use_cache"), "attention_mask": attention_mask, "output_router_logits": output_router_logits, } ) return model_inputs def prepare_inputs_for_generation( self, input_ids, past_key_values=None, inputs_embeds=None, attention_mask=None, output_router_logits=False, **kwargs, ): images = kwargs.pop("images", None) audios = kwargs.pop("audios", None) _inputs = self.prepare_inputs_for_generation_original( input_ids, past_key_values=past_key_values, inputs_embeds=inputs_embeds, attention_mask=attention_mask, output_router_logits=output_router_logits, **kwargs, ) if images is not None: _inputs["images"] = images if audios is not None: _inputs["audios"] = audios return _inputs def expand2square(self, pil_img, background_color): width, height = pil_img.size if width == height: return pil_img elif width > height: result = Image.new(pil_img.mode, (width, width), background_color) result.paste(pil_img, (0, (width - height) // 2)) return result else: result = Image.new(pil_img.mode, (height, height), background_color) result.paste(pil_img, ((height - width) // 2, 0)) return result def process_images(self, images, model_cfg): vision_tower = self.get_vision_tower() if not vision_tower.is_loaded: vision_tower.load_model() image_processor = vision_tower.image_processor image_aspect_ratio = getattr(model_cfg, "image_aspect_ratio", None) new_images = [] if image_aspect_ratio == "pad": for image in images: image = self.expand2square( image, tuple(int(x * 255) for x in image_processor.image_mean) ) image = image_processor.preprocess(image, return_tensors="pt")["pixel_values"][0] new_images.append(image) else: return image_processor(images, return_tensors="pt")["pixel_values"] if all(x.shape == new_images[0].shape for x in new_images): new_images = torch.stack(new_images, dim=0) return new_images AutoConfig.register("vita-mixtral", VITAMixtralConfig) AutoModelForCausalLM.register(VITAMixtralConfig, VITAMixtralForCausalLM)